Multiple PageRank Versions
Google has developed seven different types of PageRank, such as the deprecated nearest-seed PageRank (pageRank_NS) and the well-known ToolbarPageRank. These variations focus on different aspects, like proximity and topic relevance.
Business Models and Site Authority
Google identifies different business models (e.g., news, personal blogs, e-commerce), though it is unclear why personal blogs are specifically filtered. Contrary to past statements, the leak confirms Google does track site-wide authority metrics, including data collected via Chrome browser traffic.
Content Embedding and Topic Scores
Google’s algorithm uses page embeddings and site embeddings to evaluate how well content fits into a site’s topical focus. The siteFocusScore measures a website’s concentration on a topic, while siteRadius evaluates deviations from this topic. Consistency within these embeddings improves SEO scores.
Clicks and User Behavior Data
User engagement is heavily measured through clicks—Google tracks good, bad, and last-longest clicks. NavBoost is a mechanism that re-ranks search results based on user behavior, despite Google’s public denials of click-based rankings.
Page and Image Quality
A new metric, pageQuality (PQ), uses a large language model (LLM) to measure the “effort” in content creation. The presence of videos, tools, and images boosts PQ scores. Similarly, image quality signals are tracked based on user clicks.
Authority and Election Signals
The documents reference a factor labeled isElectionAuthority, potentially confirming the use of seed sites or topical authority sites with high influence in specific domains.
Content Freshness and Updates
Links from newer pages are favored over those embedded in older content. Google stores up to 20 versions of a webpage and retains the most recent versions for ranking. Regular updates to content are crucial to maintaining visibility in search results.
